We will learn about the following topics in this section.
Firstly, we will learn how to load a real-life dataset into R. These days using a group of packages called “tidyverse” github blog is the standard. Install tidyverse if you have never used it.
install.packages("tidyverse", dependencies = TRUE)Loading a package (and related packages) is the following.
library(tidyverse)## Loading tidyverse: ggplot2
## Loading tidyverse: tibble
## Loading tidyverse: tidyr
## Loading tidyverse: readr
## Loading tidyverse: purrr
## Loading tidyverse: dplyr## Conflicts with tidy packages ----------------------------------------------## filter(): dplyr, stats
## lag(): dplyr, statsLoad a dataset in the current working directory. You need to explicitly assign it a name.
## CSV file
framingham <- read_csv("./framingham.csv")## Parsed with column specification:
## cols(
## .default = col_integer(),
## SYSBP = col_double(),
## DIABP = col_double(),
## BMI = col_double(),
## TIMEAP = col_double(),
## TIMEMI = col_double(),
## TIMEMIFC = col_double(),
## TIMECHD = col_double(),
## TIMESTRK = col_double(),
## TIMECVD = col_double(),
## TIMEDTH = col_double(),
## TIMEHYP = col_double()
## )## See spec(...) for full column specifications.## Excel xlsx file
framinghamxl <- readxl::read_xlsx("./framingham.xlsx")Evaluating the dataset name will give you a snapshot.
framingham## # A tibble: 4,434 x 37
## SEX TOTCHOL AGE SYSBP DIABP CURSMOKE CIGPDAY BMI DIABETES BPMEDS
## <int> <int> <int> <dbl> <dbl> <int> <int> <dbl> <int> <int>
## 1 1 195 39 106.0 70 0 0 26.97 0 0
## 2 0 250 46 121.0 81 0 0 28.73 0 0
## 3 1 245 48 127.5 80 1 20 25.34 0 0
## 4 0 225 61 150.0 95 1 30 28.58 0 0
## 5 0 285 46 130.0 84 1 23 23.10 0 0
## 6 0 228 43 180.0 110 0 0 30.30 0 0
## 7 0 205 63 138.0 71 0 0 33.11 0 0
## 8 0 313 45 100.0 71 1 20 21.68 0 0
## 9 1 260 52 141.5 89 0 0 26.36 0 0
## 10 1 225 43 162.0 107 1 30 23.61 0 0
## # ... with 4,424 more rows, and 27 more variables: HEARTRTE <int>,
## # GLUCOSE <int>, PREVCHD <int>, PREVAP <int>, PREVMI <int>,
## # PREVSTRK <int>, PREVHYP <int>, DEATH <int>, ANGINA <int>,
## # HOSPMI <int>, MI_FCHD <int>, ANYCHD <int>, STROKE <int>, CVD <int>,
## # HYPERTEN <int>, TIMEAP <dbl>, TIMEMI <dbl>, TIMEMIFC <dbl>,
## # TIMECHD <dbl>, TIMESTRK <dbl>, TIMECVD <dbl>, TIMEDTH <dbl>,
## # TIMEHYP <dbl>, bmicat <int>, agecat <int>, highbp <int>, packs <int>If you want to see the all the columns.
print(framingham, width = Inf)## # A tibble: 4,434 x 37
## SEX TOTCHOL AGE SYSBP DIABP CURSMOKE CIGPDAY BMI DIABETES BPMEDS HEARTRTE GLUCOSE PREVCHD PREVAP PREVMI PREVSTRK PREVHYP DEATH ANGINA HOSPMI MI_FCHD ANYCHD STROKE CVD HYPERTEN TIMEAP TIMEMI TIMEMIFC TIMECHD TIMESTRK TIMECVD TIMEDTH TIMEHYP bmicat agecat highbp packs
## <int> <int> <int> <dbl> <dbl> <int> <int> <dbl> <int> <int> <int> <int> <int> <int> <int> <int> <int> <int> <int> <int> <int> <int> <int> <int> <int> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <int> <int> <int> <int>
## 1 1 195 39 106.0 70 0 0 26.97 0 0 80 77 0 0 0 0 0 0 0 1 1 1 0 1 0 24.000000 17.626283 17.626283 17.626283 24.00000 17.62628 24.000000 24.000000 3 1 0 0
## 2 0 250 46 121.0 81 0 0 28.73 0 0 95 76 0 0 0 0 0 0 0 0 0 0 0 0 0 24.000000 24.000000 24.000000 24.000000 24.00000 24.00000 24.000000 24.000000 3 2 0 0
## 3 1 245 48 127.5 80 1 20 25.34 0 0 75 70 0 0 0 0 0 0 0 0 0 0 0 0 0 24.000000 24.000000 24.000000 24.000000 24.00000 24.00000 24.000000 24.000000 3 2 0 1
## 4 0 225 61 150.0 95 1 30 28.58 0 0 65 103 0 0 0 0 1 1 0 0 0 0 1 1 1 8.093087 8.093087 8.093087 8.093087 5.71937 5.71937 8.093087 0.000000 3 4 1 2
## 5 0 285 46 130.0 84 1 23 23.10 0 0 85 85 0 0 0 0 0 0 0 0 0 0 0 0 1 24.000000 24.000000 24.000000 24.000000 24.00000 24.00000 24.000000 11.731691 2 2 0 2
## 6 0 228 43 180.0 110 0 0 30.30 0 0 77 99 0 0 0 0 1 0 0 0 1 1 0 1 1 24.000000 24.000000 15.657769 15.657769 24.00000 15.65777 24.000000 0.000000 4 2 1 0
## 7 0 205 63 138.0 71 0 0 33.11 0 0 60 85 0 0 0 0 0 0 1 0 0 1 0 0 1 1.021218 24.000000 24.000000 1.021218 24.00000 24.00000 24.000000 6.056126 4 4 0 0
## 8 0 313 45 100.0 71 1 20 21.68 0 0 79 78 0 0 0 0 0 0 0 0 0 0 0 0 1 24.000000 24.000000 24.000000 24.000000 24.00000 24.00000 24.000000 23.761807 2 2 0 1
## 9 1 260 52 141.5 89 0 0 26.36 0 0 76 79 0 0 0 0 1 0 0 0 0 0 0 0 1 24.000000 24.000000 24.000000 24.000000 24.00000 24.00000 24.000000 0.000000 3 3 1 0
## 10 1 225 43 162.0 107 1 30 23.61 0 0 93 88 0 0 0 0 1 0 0 0 0 0 0 0 1 24.000000 24.000000 24.000000 24.000000 24.00000 24.00000 24.000000 0.000000 2 2 1 2
## # ... with 4,424 more rowsA “tidy” dataset (data frame) should have rows as observations and columns as variables like this dataset. glimpse() gives may be more useful in providing over view of all variables.
glimpse(framingham)## Observations: 4,434
## Variables: 37
## $ SEX <int> 1, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, 0, 1, 0,...
## $ TOTCHOL <int> 195, 250, 245, 225, 285, 228, 205, 313, 260, 225, 254...
## $ AGE <int> 39, 46, 48, 61, 46, 43, 63, 45, 52, 43, 50, 43, 46, 4...
## $ SYSBP <dbl> 106.0, 121.0, 127.5, 150.0, 130.0, 180.0, 138.0, 100....
## $ DIABP <dbl> 70.0, 81.0, 80.0, 95.0, 84.0, 110.0, 71.0, 71.0, 89.0...
## $ CURSMOKE <int> 0, 0, 1, 1, 1, 0, 0, 1, 0, 1, 0, 0, 1, 0, 1, 1, 1, 1,...
## $ CIGPDAY <int> 0, 0, 20, 30, 23, 0, 0, 20, 0, 30, 0, 0, 15, 0, 9, 20...
## $ BMI <dbl> 26.97, 28.73, 25.34, 28.58, 23.10, 30.30, 33.11, 21.6...
## $ DIABETES <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
## $ BPMEDS <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0,...
## $ HEARTRTE <int> 80, 95, 75, 65, 85, 77, 60, 79, 76, 93, 75, 72, 98, 6...
## $ GLUCOSE <int> 77, 76, 70, 103, 85, 99, 85, 78, 79, 88, 76, 61, 64, ...
## $ PREVCHD <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
## $ PREVAP <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
## $ PREVMI <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
## $ PREVSTRK <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
## $ PREVHYP <int> 0, 0, 0, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 1, 1, 0,...
## $ DEATH <int> 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 1,...
## $ ANGINA <int> 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
## $ HOSPMI <int> 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
## $ MI_FCHD <int> 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
## $ ANYCHD <int> 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
## $ STROKE <int> 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
## $ CVD <int> 1, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
## $ HYPERTEN <int> 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 0,...
## $ TIMEAP <dbl> 24.0000000, 24.0000000, 24.0000000, 8.0930869, 24.000...
## $ TIMEMI <dbl> 17.6262834, 24.0000000, 24.0000000, 8.0930869, 24.000...
## $ TIMEMIFC <dbl> 17.6262834, 24.0000000, 24.0000000, 8.0930869, 24.000...
## $ TIMECHD <dbl> 17.6262834, 24.0000000, 24.0000000, 8.0930869, 24.000...
## $ TIMESTRK <dbl> 24.0000000, 24.0000000, 24.0000000, 5.7193703, 24.000...
## $ TIMECVD <dbl> 17.6262834, 24.0000000, 24.0000000, 5.7193703, 24.000...
## $ TIMEDTH <dbl> 24.0000000, 24.0000000, 24.0000000, 8.0930869, 24.000...
## $ TIMEHYP <dbl> 24.000000, 24.000000, 24.000000, 0.000000, 11.731691,...
## $ bmicat <int> 3, 3, 3, 3, 2, 4, 4, 2, 3, 2, 2, 3, 3, 4, 2, 2, 2, 2,...
## $ agecat <int> 1, 2, 2, 4, 2, 2, 4, 2, 3, 2, 2, 2, 2, 2, 1, 1, 2, 2,...
## $ highbp <int> 0, 0, 0, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0, 0, 1, 1, 0,...
## $ packs <int> 0, 0, 1, 2, 2, 0, 0, 1, 0, 2, 0, 0, 1, 0, 1, 1, 1, 1,...Various ways exist to access elements in a data frame. To access a single variable, the variable name or a position can be used. Indexing starts at 1 like in math, unlike in many programming languages.
## Single variable access. This forms a vector.
framingham$SEX
## This is the same.
framingham[["SEX"]]
## This is the same.
framingham[[1]]
## Using dimension. This forms a smaller data frame in tidyverse.
framingham[1:2, c("SEX","AGE")]We saw a vector coming out of a data frame in the previous part. An “atomic vector” is the smallest data object in R that holds values of the same type. A “list” is a heterogeneous data object that can contain different types of objects including vectors or lists. This tweet may give a better idea.
A data frame is in fact a heterogeneous list of vectors of different types (each holding a single type of data) that are all of the same length (sample size n).
An atomic vector can be of several types including.
Missing values are very common in real-life data analysis. R has a build-in way to handle this by “NA”. NA is a value that is not known. It acts like a placeholder, but gives NA for most operations.
NA + 1## [1] NANA * 1## [1] NANA + NA## [1] NANA == 1## [1] NA## Cannot detect NA by equality!
NA == NA## [1] NA## Use is.na()
is.na(NA)## [1] TRUE## For some reason, %in% (member of operation) gives FALSE.
NA %in% c(1)## [1] FALSECategorical variables are handled as “factors” in R. Factors are really integers with a finite number of levels that are labeled. In our dataset, bmicat has the following coding. They can be handy when plotting data or conducting regression analyses.
## BMI category is coded 1,2,3,4
table(framingham$bmicat)##
## 1 2 3 4
## 57 1936 1845 577## factor() can be used to associate levels to labels.
framingham$bmicat2 <- factor(framingham$bmicat,
levels = c(1,2,3,4),
labels = c("underweight","normal","overweight","obese"))
table(framingham$bmicat2)##
## underweight normal overweight obese
## 57 1936 1845 577## Change reference level to normal.
framingham$bmicat3 <- forcats::fct_relevel(framingham$bmicat2, "normal")
table(framingham$bmicat3)##
## normal underweight overweight obese
## 1936 57 1845 577## Recode by collapsing normal and underweight together with a new name.
framingham$bmicat4 <- forcats::fct_recode(framingham$bmicat2,
"normal or underweight" = "normal",
"normal or underweight" = "underweight")
table(framingham$bmicat4)##
## normal or underweight overweight obese
## 1993 1845 577Columns can be select()ed. Note the change in dimension
## Select a few variables and drop others
select(framingham, SEX, AGE, TOTCHOL)## # A tibble: 4,434 x 3
## SEX AGE TOTCHOL
## <int> <int> <int>
## 1 1 39 195
## 2 0 46 250
## 3 1 48 245
## 4 0 61 225
## 5 0 46 285
## 6 0 43 228
## 7 0 63 205
## 8 0 45 313
## 9 1 52 260
## 10 1 43 225
## # ... with 4,424 more rows## Move some of the variables to the left.
select(framingham, SYSBP, DIABP, everything())## # A tibble: 4,434 x 40
## SYSBP DIABP SEX TOTCHOL AGE CURSMOKE CIGPDAY BMI DIABETES BPMEDS
## <dbl> <dbl> <int> <int> <int> <int> <int> <dbl> <int> <int>
## 1 106.0 70 1 195 39 0 0 26.97 0 0
## 2 121.0 81 0 250 46 0 0 28.73 0 0
## 3 127.5 80 1 245 48 1 20 25.34 0 0
## 4 150.0 95 0 225 61 1 30 28.58 0 0
## 5 130.0 84 0 285 46 1 23 23.10 0 0
## 6 180.0 110 0 228 43 0 0 30.30 0 0
## 7 138.0 71 0 205 63 0 0 33.11 0 0
## 8 100.0 71 0 313 45 1 20 21.68 0 0
## 9 141.5 89 1 260 52 0 0 26.36 0 0
## 10 162.0 107 1 225 43 1 30 23.61 0 0
## # ... with 4,424 more rows, and 30 more variables: HEARTRTE <int>,
## # GLUCOSE <int>, PREVCHD <int>, PREVAP <int>, PREVMI <int>,
## # PREVSTRK <int>, PREVHYP <int>, DEATH <int>, ANGINA <int>,
## # HOSPMI <int>, MI_FCHD <int>, ANYCHD <int>, STROKE <int>, CVD <int>,
## # HYPERTEN <int>, TIMEAP <dbl>, TIMEMI <dbl>, TIMEMIFC <dbl>,
## # TIMECHD <dbl>, TIMESTRK <dbl>, TIMECVD <dbl>, TIMEDTH <dbl>,
## # TIMEHYP <dbl>, bmicat <int>, agecat <int>, highbp <int>, packs <int>,
## # bmicat2 <fctr>, bmicat3 <fctr>, bmicat4 <fctr>Rows can be filter()ed based on conditions on variables. Note the change in dimension. R does not modify your data unless you explicitly assign it back.
## Data frame name is followed by conditions that return TRUE/FALSE
filter(framingham, AGE >= 65)## # A tibble: 203 x 40
## SEX TOTCHOL AGE SYSBP DIABP CURSMOKE CIGPDAY BMI DIABETES BPMEDS
## <int> <int> <int> <dbl> <dbl> <int> <int> <dbl> <int> <int>
## 1 0 252 65 179.5 114.0 0 0 30.47 0 0
## 2 0 254 67 157.0 89.0 0 0 24.25 0 0
## 3 0 311 66 154.0 80.0 0 0 28.55 0 0
## 4 0 193 65 123.0 76.5 0 0 29.33 0 0
## 5 0 278 66 187.0 88.0 0 0 40.52 0 0
## 6 0 264 67 139.0 80.0 0 0 25.75 0 0
## 7 1 288 66 109.0 71.0 0 0 29.29 0 0
## 8 0 214 66 212.0 104.0 0 0 25.32 0 0
## 9 0 259 67 151.0 101.0 0 0 21.67 0 0
## 10 1 257 67 125.0 67.5 0 0 25.95 0 0
## # ... with 193 more rows, and 30 more variables: HEARTRTE <int>,
## # GLUCOSE <int>, PREVCHD <int>, PREVAP <int>, PREVMI <int>,
## # PREVSTRK <int>, PREVHYP <int>, DEATH <int>, ANGINA <int>,
## # HOSPMI <int>, MI_FCHD <int>, ANYCHD <int>, STROKE <int>, CVD <int>,
## # HYPERTEN <int>, TIMEAP <dbl>, TIMEMI <dbl>, TIMEMIFC <dbl>,
## # TIMECHD <dbl>, TIMESTRK <dbl>, TIMECVD <dbl>, TIMEDTH <dbl>,
## # TIMEHYP <dbl>, bmicat <int>, agecat <int>, highbp <int>, packs <int>,
## # bmicat2 <fctr>, bmicat3 <fctr>, bmicat4 <fctr>## Multiple conditions are intersected (AND)
filter(framingham, AGE >= 65, AGE < 70, SEX == 1)## # A tibble: 92 x 40
## SEX TOTCHOL AGE SYSBP DIABP CURSMOKE CIGPDAY BMI DIABETES BPMEDS
## <int> <int> <int> <dbl> <dbl> <int> <int> <dbl> <int> <int>
## 1 1 288 66 109.0 71.0 0 0 29.29 0 0
## 2 1 257 67 125.0 67.5 0 0 25.95 0 0
## 3 1 207 65 139.0 88.0 0 0 24.04 0 0
## 4 1 219 65 148.0 90.0 1 15 29.35 0 0
## 5 1 164 68 142.0 85.0 0 0 30.28 1 0
## 6 1 203 67 122.0 74.0 0 0 15.54 0 0
## 7 1 250 68 109.0 73.0 1 10 24.68 0 0
## 8 1 276 66 159.0 82.0 0 0 31.42 0 0
## 9 1 NA 65 152.5 97.5 0 0 28.35 0 0
## 10 1 214 67 127.5 80.0 0 0 22.11 0 0
## # ... with 82 more rows, and 30 more variables: HEARTRTE <int>,
## # GLUCOSE <int>, PREVCHD <int>, PREVAP <int>, PREVMI <int>,
## # PREVSTRK <int>, PREVHYP <int>, DEATH <int>, ANGINA <int>,
## # HOSPMI <int>, MI_FCHD <int>, ANYCHD <int>, STROKE <int>, CVD <int>,
## # HYPERTEN <int>, TIMEAP <dbl>, TIMEMI <dbl>, TIMEMIFC <dbl>,
## # TIMECHD <dbl>, TIMESTRK <dbl>, TIMECVD <dbl>, TIMEDTH <dbl>,
## # TIMEHYP <dbl>, bmicat <int>, agecat <int>, highbp <int>, packs <int>,
## # bmicat2 <fctr>, bmicat3 <fctr>, bmicat4 <fctr>## To actually keep the new dataset, you need to assign it.
framingham_sub <- filter(framingham, AGE >= 65, AGE < 70, SEX == 1)
framingham_sub## # A tibble: 92 x 40
## SEX TOTCHOL AGE SYSBP DIABP CURSMOKE CIGPDAY BMI DIABETES BPMEDS
## <int> <int> <int> <dbl> <dbl> <int> <int> <dbl> <int> <int>
## 1 1 288 66 109.0 71.0 0 0 29.29 0 0
## 2 1 257 67 125.0 67.5 0 0 25.95 0 0
## 3 1 207 65 139.0 88.0 0 0 24.04 0 0
## 4 1 219 65 148.0 90.0 1 15 29.35 0 0
## 5 1 164 68 142.0 85.0 0 0 30.28 1 0
## 6 1 203 67 122.0 74.0 0 0 15.54 0 0
## 7 1 250 68 109.0 73.0 1 10 24.68 0 0
## 8 1 276 66 159.0 82.0 0 0 31.42 0 0
## 9 1 NA 65 152.5 97.5 0 0 28.35 0 0
## 10 1 214 67 127.5 80.0 0 0 22.11 0 0
## # ... with 82 more rows, and 30 more variables: HEARTRTE <int>,
## # GLUCOSE <int>, PREVCHD <int>, PREVAP <int>, PREVMI <int>,
## # PREVSTRK <int>, PREVHYP <int>, DEATH <int>, ANGINA <int>,
## # HOSPMI <int>, MI_FCHD <int>, ANYCHD <int>, STROKE <int>, CVD <int>,
## # HYPERTEN <int>, TIMEAP <dbl>, TIMEMI <dbl>, TIMEMIFC <dbl>,
## # TIMECHD <dbl>, TIMESTRK <dbl>, TIMECVD <dbl>, TIMEDTH <dbl>,
## # TIMEHYP <dbl>, bmicat <int>, agecat <int>, highbp <int>, packs <int>,
## # bmicat2 <fctr>, bmicat3 <fctr>, bmicat4 <fctr>Use rename() to change variables names. If you don’t assign it back explicitly, the original data frame does not change.
## rename() can be used to rename specific variables.
## Again assignment is necessary to keep it.
rename(framingham,
total_cholesterol = TOTCHOL,
cigarretes_per_day = CIGPDAY)## # A tibble: 4,434 x 40
## SEX total_cholesterol AGE SYSBP DIABP CURSMOKE cigarretes_per_day
## <int> <int> <int> <dbl> <dbl> <int> <int>
## 1 1 195 39 106.0 70 0 0
## 2 0 250 46 121.0 81 0 0
## 3 1 245 48 127.5 80 1 20
## 4 0 225 61 150.0 95 1 30
## 5 0 285 46 130.0 84 1 23
## 6 0 228 43 180.0 110 0 0
## 7 0 205 63 138.0 71 0 0
## 8 0 313 45 100.0 71 1 20
## 9 1 260 52 141.5 89 0 0
## 10 1 225 43 162.0 107 1 30
## # ... with 4,424 more rows, and 33 more variables: BMI <dbl>,
## # DIABETES <int>, BPMEDS <int>, HEARTRTE <int>, GLUCOSE <int>,
## # PREVCHD <int>, PREVAP <int>, PREVMI <int>, PREVSTRK <int>,
## # PREVHYP <int>, DEATH <int>, ANGINA <int>, HOSPMI <int>, MI_FCHD <int>,
## # ANYCHD <int>, STROKE <int>, CVD <int>, HYPERTEN <int>, TIMEAP <dbl>,
## # TIMEMI <dbl>, TIMEMIFC <dbl>, TIMECHD <dbl>, TIMESTRK <dbl>,
## # TIMECVD <dbl>, TIMEDTH <dbl>, TIMEHYP <dbl>, bmicat <int>,
## # agecat <int>, highbp <int>, packs <int>, bmicat2 <fctr>,
## # bmicat3 <fctr>, bmicat4 <fctr>## Give a variable less confusing name and assign back to the same name.
framingham <- rename(framingham,
MALE = SEX)
framingham## # A tibble: 4,434 x 40
## MALE TOTCHOL AGE SYSBP DIABP CURSMOKE CIGPDAY BMI DIABETES BPMEDS
## <int> <int> <int> <dbl> <dbl> <int> <int> <dbl> <int> <int>
## 1 1 195 39 106.0 70 0 0 26.97 0 0
## 2 0 250 46 121.0 81 0 0 28.73 0 0
## 3 1 245 48 127.5 80 1 20 25.34 0 0
## 4 0 225 61 150.0 95 1 30 28.58 0 0
## 5 0 285 46 130.0 84 1 23 23.10 0 0
## 6 0 228 43 180.0 110 0 0 30.30 0 0
## 7 0 205 63 138.0 71 0 0 33.11 0 0
## 8 0 313 45 100.0 71 1 20 21.68 0 0
## 9 1 260 52 141.5 89 0 0 26.36 0 0
## 10 1 225 43 162.0 107 1 30 23.61 0 0
## # ... with 4,424 more rows, and 30 more variables: HEARTRTE <int>,
## # GLUCOSE <int>, PREVCHD <int>, PREVAP <int>, PREVMI <int>,
## # PREVSTRK <int>, PREVHYP <int>, DEATH <int>, ANGINA <int>,
## # HOSPMI <int>, MI_FCHD <int>, ANYCHD <int>, STROKE <int>, CVD <int>,
## # HYPERTEN <int>, TIMEAP <dbl>, TIMEMI <dbl>, TIMEMIFC <dbl>,
## # TIMECHD <dbl>, TIMESTRK <dbl>, TIMECVD <dbl>, TIMEDTH <dbl>,
## # TIMEHYP <dbl>, bmicat <int>, agecat <int>, highbp <int>, packs <int>,
## # bmicat2 <fctr>, bmicat3 <fctr>, bmicat4 <fctr>To create transformed variables, mutate() is used. Again if you don’t assign explicitly, the original data frame does not change despite the name mutate().
framingham <- mutate(framingham,
htn = as.numeric(SYSBP > 140 | DIABP > 90))We often have to apply these functions sequentially. This can be done by nesting functions or chaining them with the pipe operator %>%. Here is an example where I change variable name, then filter() based on age and sex (male only), and then select() several variables, and sort by AGE. See the similarity and differences. Again the original framingham dataset is not modified unless you explicitly assign it back.
## Nesting can be ugly.
arrange(select(filter(rename(framingham,
total_cholesterol = TOTCHOL),
AGE >= 65, AGE < 70, MALE == 1),
MALE, AGE, total_cholesterol),
AGE)## # A tibble: 92 x 3
## MALE AGE total_cholesterol
## <int> <int> <int>
## 1 1 65 207
## 2 1 65 219
## 3 1 65 NA
## 4 1 65 251
## 5 1 65 285
## 6 1 65 171
## 7 1 65 225
## 8 1 65 201
## 9 1 65 167
## 10 1 65 193
## # ... with 82 more rows## Pipes to the rescue.
framingham %>%
rename(total_cholesterol = TOTCHOL) %>%
filter(AGE >= 65, AGE < 70, MALE == 1) %>%
select(MALE, AGE, total_cholesterol) %>%
arrange(AGE)## # A tibble: 92 x 3
## MALE AGE total_cholesterol
## <int> <int> <int>
## 1 1 65 207
## 2 1 65 219
## 3 1 65 NA
## 4 1 65 251
## 5 1 65 285
## 6 1 65 171
## 7 1 65 225
## 8 1 65 201
## 9 1 65 167
## 10 1 65 193
## # ... with 82 more rowsWe sometimes want to examine summary statistics, often within some subgroups. Here we construct subgroups by sex and previous history of coronary heart disease. na.rm = TRUE means drop the missing values NA when calculating the mean, sd or variance (complete case analysis).
framingham %>%
group_by(MALE, PREVCHD) %>%
summarize(mean_total_chol = mean(TOTCHOL, na.rm = TRUE),
sd_total_chol = sd(TOTCHOL, na.rm = TRUE),
var_total_chol = var(TOTCHOL, na.rm = TRUE))## # A tibble: 4 x 5
## # Groups: MALE [?]
## MALE PREVCHD mean_total_chol sd_total_chol var_total_chol
## <int> <int> <dbl> <dbl> <dbl>
## 1 0 0 239.4413 46.19686 2134.150
## 2 0 1 248.0735 46.57663 2169.383
## 3 1 0 233.1048 42.13729 1775.551
## 4 1 1 240.5242 45.04297 2028.869For the purpose of “Table 1” (baseline patient characteristics), a specialized package may be more useful.
library(tableone)
mytab1 <- CreateTableOne(vars = c("TOTCHOL","AGE","BMI","DIABETES","BPMEDS"), # Variables to include
strata = c("MALE", "PREVCHD"), # Classification variable
data = framingham, # dataset
factorVars = c("DIABETES","BPMEDS")) # variables to handle as categoricals.
print(mytab1)## Stratified by MALE:PREVCHD
## 0:0 1:0 0:1
## n 2420 1820 70
## TOTCHOL (mean (sd)) 239.44 (46.20) 233.10 (42.14) 248.07 (46.58)
## AGE (mean (sd)) 49.80 (8.60) 49.29 (8.53) 58.29 (5.68)
## BMI (mean (sd)) 25.51 (4.50) 26.19 (3.42) 28.42 (5.69)
## DIABETES = 1 (%) 57 (2.4) 52 (2.9) 5 ( 7.1)
## BPMEDS = 1 (%) 89 (3.7) 35 (1.9) 13 (18.8)
## Stratified by MALE:PREVCHD
## 1:1 p test
## n 124
## TOTCHOL (mean (sd)) 240.52 (45.04) <0.001
## AGE (mean (sd)) 57.02 (8.23) <0.001
## BMI (mean (sd)) 25.94 (3.27) <0.001
## DIABETES = 1 (%) 7 (5.6) 0.015
## BPMEDS = 1 (%) 7 (6.0) <0.001